Vision-Based Weld Pool Width Control

1994 ◽  
Vol 116 (1) ◽  
pp. 86-92 ◽  
Author(s):  
K. A. Pietrzak ◽  
S. M. Packer
Keyword(s):  
Control System ◽  
Weld Pool ◽  
Majority Voting ◽  
Weld Penetration ◽  
Stainless Steel 304 ◽  
Weld Pool Geometry ◽  
Vision Algorithm ◽  
Welding Processes ◽  
Weld Pool Width ◽  
Matched Edge

Methods for controlling weld penetration for arc welding processes from top-side measurements have long been sought. One indirect variable that has been reported to correlate with penetration is weld pool geometry. A system which uses weld pool geometry sensing for controlling weld penetration is described in this paper. The system uses a miniature camera mounted in a modified coaxial viewing torch to view the weld pool. A robust machine vision algorithm has been developed for this system to measure weld pool width. The algorithm was designed to locate the edges of the weld pool despite the presence of other edges caused by the heat affected zone, scratches, marks, and weld pool impurities. The algorithm uses a matched edge filter and a majority voting scheme to measure the width of the pool. A control system was developed to regulate weld pool width in the presence of disturbances caused by such items as incorrect parameter settings, small variations in material composition, and material thickness changes. Experiments were conducted to test the control system by simulating some of these disturbances. The experiments demonstrated that for certain classes of materials, this technique works quite well. However, for other materials such as stainless steel 304, surface impurities in the weld pool visually obscure the weld pool and its edges to such a degree that the system fails to lock onto the edges of the pool.

Advanced Scaling Techniques for the Modeling of Materials Processing

2009 ◽  
Author(s):  
Karem Tello ◽  
Ustun Duman ◽  
Patricio Mendez
Keyword(s):  
Weld Pool ◽  
Arc Welding ◽  
Maximum Temperature ◽  
Scaling Analysis ◽  
Weld Penetration ◽  
Friction Stir ◽  
Gas Tungsten Arc ◽  
Pool Surface ◽  
Scaling Models ◽  
Welding Processes

The present work presents how scaling analysis can be applied into multiphysics and multicoupled problems related to welding processes. The formation of the weld pool surface depression in high current and velocity Gas Tungsten Arc Welding (GTAW) is dominated by the gas shear acting on the weld pool. Considering this dominant force the weld penetration was estimated and compared to experimental results. Plastic deformation and heat flow are coupled phenomena in Friction Stir Welding (FSW), the maximum temperature was estimated using scaling analysis and compared with experimental and numerical results reported in the literature. Although the simplicity of the scaling models, they are capable of capturing correct trends and order of magnitudes of the unknown estimations in a problem. Moreover, they are capable of determining the dominant forces that act on the process studied.

scholarly journals Deep Learning-Based Detection of Penetration from Weld Pool Reflection Images

2020 ◽  
Vol 99 (9) ◽  
pp. 239s-245s
Author(s):  
CHAO LI ◽  
◽  
QIYUE WANG ◽  
WENHUA JIAO ◽  
MICHAEL JOHNSON ◽  
...  
Keyword(s):  
Weld Pool ◽  
Data Augmentation ◽  
Welding Process ◽  
Descent Method ◽  
Machine Learning Techniques ◽  
Gradient Descent Method ◽  
Neural Net ◽  
Weld Penetration ◽  
Process Data ◽  
Pool Surface

An innovative method was proposed to determine weld joint penetration using machine learning techniques. In our approach, the dot-structured laser images reflected from an oscillating weld pool surface were captured. Experienced welders typically evaluate the weld penetration status based on this reflected laser pattern. To overcome the challenges in identifying features and accurately processing the images using conventional machine vision algorithms, we proposed the use the raw images without any processing as the input to a convolutional neural network (CNN). The labels needed to train the CNN were the measured weld penetration states, obtained from the images on the backside of the workpiece as a set of discrete weld penetration categories. The raw data, images, and penetration state were generated from extensive experiments using an automated robotic gas tungsten arc welding process. Data augmentation was performed to enhance the robustness of the trained network, which led to 270,000 training examples, 45,000 validation examples, and 45,000 test examples. A six-layer convolutional neural net-work trained with a modified mini-batch gradient descent method led to a final testing accuracy of 90.7%. A voting mechanism based on three continuous images increased the classification accuracy to 97.6%.

Estimation of Weld Pool Sizes in GMA Welding Process Using Neural Networks

1993 ◽  
Vol 207 (1) ◽  
pp. 15-26 ◽  
Author(s):  
T G Lim ◽  
H S Cho
Keyword(s):  
Neural Network ◽  
Weld Pool ◽  
Gma Welding ◽  
Quality Monitoring ◽  
Design Parameters ◽  
Geometrical Parameters ◽  
The Neural Network ◽  
On Line ◽  
Welding Processes ◽  
Pool Sizes

In gas metal arc (GMA) welding processes, the geometrical shape and size of the weld pool are utilized to assess the integrity of the weld quality. Monitoring of these geometrical parameters for on-line process control as well as for on-line quality evaluation, however, is an extremely difficult problem. The paper describes the design of a neural network estimator to estimate weld pool sizes for on-line use in quality monitoring and control. The neural network estimator is designed to estimate the weld pool sizes from surface temperatures measured at various points on the top surface of the weldment. The main task of the neural network is to realize the mapping characteristics from the point temperatures to the weld pool sizes through training. The chosen design parameters of the neural network estimator, such as the number of hidden layers and the number of nodes in a layer, are based on an estimation error analysis. A series of bead-on-plate welding experiments were performed to assess the performance of the neural network estimator. The experimental results show that the proposed neural network estimator can estimate the weld pool sizes with satisfactory accuracy.

Transport Phenomena and Their Effect on Weld Quality in GMA Welding of Aluminum Alloys

Volume 3 ◽  
2004 ◽  
Author(s):  
H. Guo ◽  
H. L. Tsai ◽  
P. C. Wang
Keyword(s):  
Aluminum Alloys ◽  
Weld Pool ◽  
Gma Welding ◽  
Gas Metal Arc Welding ◽  
Welding Current ◽  
Weld Penetration ◽  
Three Dimensional Model ◽  
End Effects ◽  
Welding Arc ◽  
Micro Cracks

Gas metal arc welding (GMAW) of aluminum alloys has recently become popular in the auto industry to increase fuel efficiency of a vehicle. In many situations, the weld is short (say, less than two inches) and the “end effects” become very critical in determining the strength of the weld. At the beginning stage of the welding, when the metal is still “cold”, which is frequently called cold weld, limited weld penetration occurs. On the other hand, at the ending stage of the welding, a “crater” is formed involving micro-cracks and micro-pores. Both the cold weld and the crater can significantly decrease the strength of the weld and are more severe for aluminum alloys as compared to steels. Hence, there are strong needs to improve the GMAW process in order to reduce or eliminate the aforementioned end effects. In this paper, both mathematical modeling and experiments have been conducted to study the beginning stage, ending stage, as well as the quasi-steady-state stage of GMA welding of aluminum alloys. In the modeling, a three-dimensional model using the volume-of-fluid (VOF) method is employed to handle the free surfaces associated with the impingement of droplets into the weld pool and the weld pool dynamics. Transient weld pool shapes and the distributions of temperature and velocity in the weld pool are calculated. The predicted solidified weld bead shapes, including weld penetration and/or reinforcement, are in agreement with experimental results for welds in the aforementioned three stages. It was found that the thickness of the molten weld pool is smaller and there is no vortex developed, as compared to steel welding. The lack of penetration in cold weld is due to the lack of pre-heating by the welding arc. Three techniques are proposed and validated numerically to improve weld penetration by increasing the energy input at the beginning stage of the welding. The crater formation is caused by rapid solidification of the weld pool when the welding arc is terminated. By reducing welding current and reversing the welding direction before terminating the arc, the weld pool is maintained “hot” for a longer time allowing melt flow to fill-up the crater. This method is validated experimentally and numerically to be able to eliminate the formation of the crater and the associated micro-cracks.

Sign in / Sign up

Export Citation Format

Share Document